October this year saw publication of Towards Open Research: Practices, experiences, barriers and opportunities, a study investigating researcher’s attitudes and behaviours in respect of open research, commissioned by the Wellcome Trust, and based on surveys of researchers and focus groups conducted by the study and the Economic and Social Research Council (ESRC).
The aim of the study was to identify practical actions the Wellcome Trust can implement to remove barriers and maximise the opportunities for practicing open science. Under the aegis of open science the report study considered Open Access publishing, data sharing and re-use, and code-sharing and re-use.
Both the Wellcome Trust and ESRC are strong proponents of open research. The Wellcome Trust has long been at the forefront of policy initiatives to advance an open science agenda through Open Access and open data practices. It mandates both Open Access publication of funded research outputs and data sharing to the fullest achievable extent, and supports these activities through its research grants. The report commissioned by the Trust coincides with the launch of Wellcome Open Research, a platform on which its funded researchers can rapidly publish any results they wish to share, including study protocols and null and negative results, as well as articles and data, which are all made available without editorial intervention for open peer review.
ESRC similarly mandates both Open Access publication and data sharing wherever possible; Open Access publishing for its funded researchers is supported through the RCUK block grant to institutions, and it manages the UK Data Archive, the UK’s largest collection of social, economic and population data and a service that its funded researchers can use to preserve and share the data arising from their research. The UK Data Archive has been in existence since 1967, and ESRC was one of the earliest among funders to adopt a data sharing policy, in the mid-1990s, making it an informed and progressive force in the promotion of open and accessible data, with particular expertise in the management of controlled access to disclosive and sensitive data. It also offers a wealth of invaluable supporting information and resources for research data management via the UK Data Service website, covering topics such as participant consent for data sharing, anonymisation of datasets, and dealing with rights in data.
The study finds that open research is widely practised and on the increase, with researchers not only using Open Access and data sharing, but engaging in growing numbers in other emerging open practices, such as code sharing and open peer review, and experiencing the benefits of increased citation rates, and accelerated communication of a broader range of research outputs beyond the traditional peer-reviewed journal paper.
- Over 70% of Wellcome-funded papers are published as Open Access and a third of researchers publish all their papers as Open Access.
- Researchers appreciate the value of Wellcome’s new Open Research platform in enabling the rapid dissemination of research materials and results, enabling data visualisation in papers, and providing a forum for open and constructive peer review of of methods and findings.
- Half of researchers make their data available for use by others, largely via institutional and community data repositories.
- The drivers for researchers to share data largely come from funder and journal requirements; although researchers generally accept the case for sharing data, very few report any direct benefits from sharing their data, and many are still concerned about possible misuse or misinterpretation of their data, loss of first-use privilege to competitors, and deterred by the effort required to prepare and deposit data.
- Early career researchers in particular may show reluctance to share their data for fear of losing future publication and career progression opportunities by releasing their research capital. Their supervisors and seniors have a role to play in encouraging good practice.
- On the plus side, very few researchers have had negative experiences from sharing their data, and many of the fears reported by researchers are largely unfounded.
- The key things that those who fund and support researchers can do to encourage sharing of data are: provide funding to cover the cost of data preparation, and create incentive systems that reward and recognise researchers for sharing data.
- Approximately three quarters of researchers have re-used research data, mostly to provide background information and context to research, for research validation, and to help develop methodologies for new analyses.
- Data sharing can be complex and effort-intensive: researchers need to be given training and guidance, to have easy-to-use data repository services, and to be supported in producing and curating high-quality data and addressing challenges such as making disclosive and confidential data safe for sharing, or sharing large resources, e.g. imaging data.
- Code-sharing is less well-established than other open practices, at least in part because fewer researchers create software code in their research: two-fifths of researchers do so (mostly those using surveys, secondary analysis and simulation), but less than half of them make it available for access and re-use by others.
- A significant amount of code use may be hidden and opportunity for code sharing not realised: some researchers may think of research code in terms of software outputs, and not consider processing scripts (such as stata.do files and batch files) within this definition, even though they may be essential to the replication and validation of research results.
- Where code-sharing takes place, it is driven far less than data sharing by the requirements of funders and journals, and more by a desire to engage in good research practice and to enable other researchers to collaborate and contribute to the work.
- There are no significant barriers to code-sharing, although lack of skills and funding and rapid changes in software can disincline researchers to invest the time and effort, especially where code sharing is not widely funded, incentivised or rewarded by funders and research organisations, and where norms of code citation and acknowledgement of re-use are not well-established. Code, unlike data in many cases, must be actively maintained and supported once it has been distributed and established a user community, and this demands both commitment from code developers and significant resources. This needs to be recognised by funders and properly supported.
- Code re-use is currently limited, with just over a third of researchers having used existing code in their research.
- Many researchers pick up software skills ad hoc, and lack formalised training or knowledge of best practice in code development and sharing. Software skills acquisition needs to be better integrated into standard researcher training models, and researchers should be able to draw on the support offered, e.g. through the Software Sustainability Institute’s Software Carpentry activities.
- While many code repositories are available for hosting and development, such as GitHub and Bitbucket, it is not clear that these are reliable long-term preservation solutions, and there may be a need for better provision of repositories dedicated to preservation and maintenance of research software. Wellcome is exploring the possibility of setting up such a code repository.
Open research in general
The overall message for funders is that they need to incentivise and reward not just the production of original and interesting research, but the whole ensemble of ‘open’ practices that together ensure the quality, accessibility and usability of all research. The communication of research outputs is not auxiliary or incidental to research, but at its very heart, and however inherently valuable a piece of research may be, its value is diminished if it is not communicated at all because it is a negative or null result, or because the researcher sees no benefit in communicating it, or it is communicated but accessible only to a few selected by ability to pay, or its methods are not transparent and open to replication and critique, or the tools it used are unavailable to others. If researchers are to be incentivised to engage with Open Research practices, then those who fund and reward research need to ensure the money is in the right places for them.
Data from surveys of 583 Wellcome Trust-funded researchers and 259 ESRC-funded researchers, and 5 focus group discussions are available from the UK Data Archive.