In 2013, this journal and many of the Nature research journals announced initiatives aimed at “reducing our irreproducibility” (Nature 496, 398; 2013). These included a life-sciences checklist for authors and editors intended to improve the transparency of the statistical and methodological aspects of laboratory work, together with abolition of length limits in online methods descriptions and greater attention to statistical evaluation.
At the same time, we encouraged the publishing of step-by-step protocols that are linked to the published papers and made
available through the open repository Protocol Exchange. And, complementing our policy of mandated deposition for certain data types, we strongly encouraged or (in some cases) mandated the provision of source data underlying graphical items.
Anecdotal feedback suggests that our application of the checklists — which represent extra time and effort by both authors and editors — has been much appreciated, although not by everybody: author compliance can be an issue, and we will soon announce steps to improve matters.
We have continued to implement policies that support reproducible research — by strengthening requirements for code availability in 2014, and introducing reporting standards for cell-line source and authentication details in 2015. A data policy, effective in 2016, introduces a mandatory data-availability statement in all papers published in the Nature journals and encourages data citation. Another notable step forward comes with the introduction of ‘registered reports’ at Nature Human Behaviour, a format intended to minimize research bias by basing acceptance on the significance of the question and the robustness of the methods, rather than the outcome of the results.
On other fronts, we have explored reproducibility-related issues in our news and opinion pages (see go.nature.com/2ca0ej1). We have also developed the checklist approach by implementing new modules for specialized areas of research afflicted by poor reporting of experimental details — in photovoltaics, laser physics and functional magnetic resonance imaging.
This week sees a further development. Nature and the Nature journals are published by Springer Nature, whose publications also include Scientific Reports, Scientific Data, Nature Partner Journals and BioMed Central and Springer journals. All of these publications are now committed to becoming formal signatories to the Transparency Openness Promotion (TOP) guidelines. The TOP guidelines (https://cos.io/top) focus on transparency and openness in research design, data and materials to enable reproducible research. They were developed with the involvement of journals (including the Nature group), and were introduced in 2015.
The guidelines consist of eight standards — citation standards, data transparency, analytic methods (code) transparency, research materials transparency, design and analysis transparency, pre-registration of studies, preregistration of analysis plans, and replication — with three levels of increasing rigour. The TOP guidelines provide a common set of standards and a useful framework for advancing an agenda for reproducible research, but uptake of individual standards by Springer Nature journals will be guided by disciplinary norms.
All of these initiatives should help those wishing to replicate our papers. The Nature journals do not have a dedicated format for replication studies, but we do consider high-value replications, subjecting them to the same criteria as other submitted studies. Scientific Data welcomes submissions describing data sets from replication studies, and recently launched an online collection highlighting a series of replication data sets it had published over the past six months. The collection was organized in partnership with the Open Science Framework, a service from the non-profit Center for Open Science in Charlottesville, Virginia, which has coordinated the development of the TOP guidelines.
As illustrated at a meeting on reproducibility issues and remedies last week at the US National Academy of Sciences, this journey is far from complete, and all of us in the research landscape are stakeholders in its progress. Nature will continue to play its part in championing the increased robustness of research.
- Journal name:
- Date published: