The availability of study data is crucial to ensure the reproducibility of research. In recent years, publisher’s submission requirements have helped to improve the transparency and quality of research reporting by encouraging data sharing, and data sharing statements are now commonplace. Public Library of Science (PLOS), a strong advocate of open science, is one publisher that has adopted policies requiring data sharing for papers they publish, believing that sharing data leads to scientific progress. Following his talk at the 2020 European Medical Writers Association symposium, The Publication Plan spoke to Iain Hrynaszkiewicz, Director, Open Research Solutions at PLOS, to find out what publishers can do to further encourage data sharing.
As an advocate for open research, what has been your experience of the shift in attitude towards data sharing in recent years?
“Data sharing policies have been in place in some communities and contexts for more than 20 years, but until around 2012, data sharing policies from journals were uncommon. Journals have tended to take incremental steps towards data sharing, starting with ‘encourages sharing’ approaches to policies. Research I was involved in carrying out showed that around 5% of researchers tend to provide a data sharing statement when these are optional. Interestingly, researchers who opt in are more likely to follow good practice, such as providing links in their published articles to data in repositories. However, sharing data alongside published articles only shows significant increases as a result of mandatory policies supported by journals. Big increases in data sharing statements, for example, have been seen in PLOS and BMC journals since mandates were introduced in 2014 and 2015, respectively.
Sharing data alongside published articles only shows significant increases as a result of mandatory policies supported by journals.
In terms of researcher attitudes, the annual State of Open Data surveys, conducted by Digital Science, provide a useful longitudinal view of researcher opinions. For example, support for funder mandates for sharing data appears to have been growing in the last few years.”
Support for funder mandates for sharing data appears to have been growing in the last few years.
What are some of the key challenges or obstacles that researchers might face that prevent or discourage them from data sharing? What can publishers do to help overcome these issues?
“Fortunately, we have quite a lot of evidence about the problems researchers have with data sharing. Many studies have highlighted problems of a lack of time, resources, skills and incentives to share data. The most commonly reported practical issues tend to be a lack of time.
But topmost among researcher concerns is usually potential misuse of their data if they make it openly available. I see this more as a trust or cultural issue than a practical one, which therefore may not be easily solved by technology or services alone.
Topmost among researcher concerns is usually potential misuse of their data if they make it openly available. I see this more as a trust or cultural issue than a practical one.
More recently, we undertook some research at PLOS to better understand how important potential barriers to data sharing are to researchers (rather than how common these problems are). We were curious why, despite many solutions being available to help with data sharing, best practice approaches – such as using data repositories – are still uncommon. We discovered that, amongst our cohort at least, while data sharing is important from the researcher’s perspective, researchers are satisfied with their current ability to share data. From this, we infer that they may be unlikely to seek new solutions (such as repositories) for sharing their data. As a result, at PLOS we see a need to focus on advocacy and education about the benefits of best practices – which include greater research impact and reliability – and seamlessly integrating existing data sharing solutions.”
Data supporting research published in manuscripts is often available ‘on request’ or included in supplementary materials. You have indicated that data repositories are the optimal method for public sharing. Should the publishing industry be doing more to promote data deposition in repositories?
“Yes, absolutely. Study after study has found that data simply ‘available on request’ are not available in practice. At PLOS we work with authors to ensure generic ‘data available on request’ statements are not an option in published articles, and we implore more journals and publishers to take a similar approach.
Study after study has found that data simply ‘available on request’ are not available in practice. At PLOS we work with authors to ensure generic ‘data available on request’ statements are not an option in published articles, and we implore more journals and publishers to take a similar approach.
Indeed, data repositories are often best practice for data sharing but, even at PLOS journals, they are not the most common method for publicly sharing data. Publishers should redouble efforts to make data sharing in repositories as effortless for researchers as sharing data in supporting information – the most popular method. Unless we do more to integrate good practice into publishing workflows, further increases in adoption will be slow.
Publishers should redouble efforts to make data sharing in repositories as effortless for researchers as sharing data in supporting information.
PLOS and other publishers often deposit supporting information files on the author’s behalf in figshare, a general repository. However, better curation and connecting of data in published articles ahead of publication, alongside detailed data availability statements, better aids data discovery and reuse in the long term.”
Data sharing has been linked with increased citations that can benefit a researcher. Should publishers do more to get this message across? Are further motivations to share data needed and if so, what do you have in mind?
“Data sharing has been repeatedly associated with more citations, yes, although this is correlation not causation. Publishers could certainly do more to promote this potential benefit, noting the caveats, as increased citations have been shown to be a motivator for sharing research data. But another, often more common and important incentive for sharing data – when researchers are surveyed – is that sharing research data will benefit both science and society. This factor is appealing to altruistic motivations but speaks more fundamentally to why we do research in the first place – which is one reason why it can’t be ignored.
Often amongst the top incentives to share research data are journal, and funder, policies. If implemented effectively, these policies are proven to work.
Also, often amongst the top incentives to share research data are journal, and funder, policies. If implemented effectively, these policies are proven to work – more data are available when policies require it, making policies an important tool in the box of available solutions to support data sharing.”
Do you think that all journals should mandate data sharing or formally review available data during the peer review process? If so, how can journals encourage peer reviewers to perform such a task?
“I am of the view that data sharing should be mandatory in journals, yes, but we must retain flexibility about how and when data are shared to acknowledge domain and cultural needs and differences in good practice. ‘As open as possible, as closed as necessary’ is a good guideline – recognising that some data, such as legally or ethically sensitive data, cannot be fully in the public domain.
‘As open as possible, as closed as necessary’ is a good guideline – recognising that some data, such as legally or ethically sensitive data, cannot be fully in the public domain.
It is possible to enable controlled access, such as for editors and reviewers during peer review, but this can be a substantial effort. Peer review of data as a mandatory requirement is only going to be practical in certain cases, such as at data journals. A practical solution that could be more widely adopted is to encourage reviewers to consider data availability in their assessment of papers, such as through adding a question to reviewer forms – an approach used at PLOS.
Peer review of data as a mandatory requirement is only going to be practical in certain cases…A practical solution that could be more widely adopted is to encourage reviewers to consider data availability in their assessment of papers.
We can also make it easier for reviewers to engage with data. At PLOS, authors’ data availability statements are visible on page 1 of the manuscript seen by reviewers, which is a simple approach to increase engagement with data and data availability during peer review.”
Last year, you published a paper ‘Developing a research data policy framework for all journals and publishers’ and called for publisher collaboration and adoption of standard data policies. What has been the feedback on this, and what further plans do you have in mind to accomplish its goals?
“We’ve been pleased to see the recommendations and policy framework in the paper reused in several contexts. This includes policies of some Springer Nature, and all PLOS, journals being updated to align with the recommendations. Also, the framework is being recommended by the STM Association as a tool to encourage their 150+ member publishers to implement data policies through collaboration with the Research Data Alliance Interest Group. These are important reuses, but seeing some more organic adoption of the framework has been particularly encouraging. A group of Slovenian journals have implemented the framework and published a paper about their experience, and Earth science and biodiversity journals have used the framework to identify improvements to their policies.
The next problem we…wish to tackle, beyond journal policy, is funder policy alignment and how funder and journal policies match up.
The next problem we – the Research Data Alliance Interest Group behind the policy framework – wish to tackle, beyond journal policy, is funder policy alignment and how funder and journal policies match up. In fact, we are initiating a collaborative project with a group of funding agencies to explore the landscape of journal and funder research data policies and the potential for alignment. I hope this leads to new recommendations and pilot initiatives in the future, as we’ve seen that policy can help achieve the goal that we all share of increasing accessibility of data to drive new discoveries.”
With thanks to our sponsor, Aspire Scientific Ltd