Data sharing in research - cultural and technical barriers in Life Sciences and Public Health - Part 2 - Barriers

The second section of our review of attitudes to data sharing in Life Sciences and Public Health was due to  look at policies introduced by research funders to oblige researchers to make their data accessible to the community, however, there is more to be said on this than we have currently committed to bytes or paper.

Thus we'll jump ahead here to our section on barriers to data-sharing and aim to get the section on funders' policies to you within the week...

Just a reminder that it's meant to serve as a 'primer' aggregating analyses from the last few years in this area and pointing you to the original articles and as such references other publications pretty heavily - so do check out the footnotes if you want to explore further.

Barriers – incentives and expertise

Discussions of this subject highlight the lack of active incentives (rather than obligations) to share data: “the lack of explicit career rewards, and in particular the perceived failure of the Research Assessment Exercise (RAE) explicitly to recognise and reward the creating and sharing of datasets – as distinct from the publication of papers - are major disincentives.”[1]

As mentioned above when looking at Social and Public Health Sciences the Research Intelligence Network found “found scant evidence of researchers wanting to publish datasets. Typically researchers will request data from one or more publicly-available datasets and they will undertake analysis. Often this process leads to the creation of new, derived datasets but these tend not to find their way to the public domain.”[2]

Lack of incentive is blamed for this outcome: “Unlike in some of the other areas we have looked at there are no obvious rewards that accrue to researchers who decide to make their datasets publicly-available – though few deny that sharing datasets produced with public funds is a worthwhile principle….. researchers producing small scale datasets see no reason to invest the time and effort required to make their datasets publicly available. Besides which, some want to control their data, limit the possibility of the data being misrepresented, and limit the scope for competition [our italics].” [3]

“Other disincentives include lack of time and resources; lack of experience and expertise in data management and in matters such as the provision of good metadata; legal and ethical constraints; lack of an appropriate archive service; and fear of exploitation or inappropriate use of the data….. Relatively few researchers have the expertise, resources and inclination to perform themselves all the tasks necessary to make their data not only available, but readily accessible and usable by others.”[4] Many researchers lack the skills to meet the quality standards imposed by data centres without substantial help from specialists.

Additionally, “creating longitudinal datasets is an expensive business and therefore the people responsible for them tend to feel the need to protect them. This is manifested in reported anxiety about commercial organisations using data, deriving slightly or materially different datasets and claiming intellectual property rights over these new datasets.”[5]

Across the biomedical sciences directors and PIs see their restricted data as their intellectual capital:  “As with most areas of research, there is competition between researchers to produce the best work in the best journal… Many researchers wish to retain exclusive use of the data they have created until they have extracted all the publication value they can.”[6]

From the perspective of commercial / industry groups data sharing presents many of the same challenges: Intellectual Property and Confidentiality are particularly sensitive issues. In the past big pharmaceuticals organisations have traditionally been conservative over data sharing – concerns include loss of control, cost, other units reaching different conclusions or deriving novel insights which may have commercial value.

Within this field there exists a significant heterogeneity of needs. The complexity and diversity of the biomedical research landscape breeds diversity in tools and methodologies for data capture and analysis, storage, maintenance and curation and this too may fuel confusion and dampen enthusiasm.

---

[1] To Share or not to Share: Publication and Quality Assurance of Research Data Outputs - - Report commissioned by the Research Information Network (RIN) in association with the Joint Information Systems Committee and the National Environment Research Council (NERC) – published June 2008. This report covered six discrete research areas, two of which were Social and Public Health Sciences and Genomics and two interdisciplinary areas, one of which was Systems Biology.

[2] Ibid.

[3] Ibid.

[4] Ibid.

[5] Ibid.

[6] Ibid.