Reproducibility of research is critical for open science and open Britain

The announcement of Plan S, which calls for full and immediate access to scientific journals (by 2020), has triggered much media debate on the principles of open access. Open research, and in particular how research reproducibility impacts science, has also been in the news, if to a lesser extent. 

Times Higher Education’s recent article discussing whether researchers can predict if individual scientific findings will replicate was an engaging take on the topic, but it’s one we need to ensure remains on the academic agenda in 2019.

Regardless of how Brexit unfolds, technological trends will continue; these include big data, social media, machine learning, robotics, and the internet of things, often collectively called the fourth industrial revolution. These trends influence and are shaped by research as much as by industry and are recognised in government policy as a vital means of driving the UK economy.

The fourth industrial revolution puts research in the spotlight; if industry and academia are to collaborate more effectively and meet the government’s target of 2.4 per cent economic growth from research and development, trust in research will be critical. 

A scientific community that can be relied upon to produce robust, reproducible findings is also one that will stimulate investment and generate economic and societal benefits.

Ahead of Brexit, and as research-intensive universities look for future research funding opportunities, the UK’s reputation for world-leading science is likely to come under increased scrutiny. If we are to attract investment from beyond the European Union and our own government, what can be done to maintain and increase trust in research?

As argued in a recent “Manifesto for Reproducible Science”, open research may play a part, by making the scientific workflow available for scrutiny, and serving as an implicit quality control process.

For example, not publishing null results is a major threat to the conclusions that we can draw from scientific research; studies that fail to observe an effect (including failures to replicate previously observed effects) are still less likely to be published.

While open access to peer-reviewed publications is important for achieving open science, this is just one part of the solution; data and study materials that underpin findings also need to be as open as possible.

Plan S principles touch upon the need to change the incentive structures for researchers, but FAIR (findable, accessible, interoperable and reusable) open data is left out of the picture. The international consortium of research funders, the cOAlition, could have used its Plan S announcement to signal that funders will soon be turning attention to FAIR and open research practices. This would help reproducibility to be considered as a part of the research lifecycle, rather than risk it being an afterthought.

To ensure reproducibility is justly considered, a group of academics recently established the UK Reproducibility Network, which has grown rapidly in just a few months. Top of the list of actions for this group is the promotion of open research practices and carrying out meta-research into how effective such initiatives are. The network will provide a platform to advocate for specific practices, in an evidence-based way, and provide training for UK researchers.

Around the globe, others are investigating how to support researchers to produce more robust evidence, such as the Center for Open Science in the US, which has links with programmes in Japan and Canada.  

Such efforts help researchers overcome hurdles in technical infrastructure that can prevent the sharing of data. The UK Reproducibility Network, Jisc, and others are building the UK’s research capabilities; contributing to the global research effort and the UK’s long tradition of meta-research.

With a  focus on technology, Jisc recently published a  report into the tools that can support robust research practices. It highlights the ongoing cultural shift towards open research through practical changes made by researchers themselves, and those adopted by publishers and funders.

There are also ways in which other sector support bodies can help. UKRI will be commissioning a roadmap for research and innovation restructure – including a strand on research data – and the work of the Open Research Data Task Force will this year make recommendations for sector bodies and universities to implement changes that will produce more open science and research.

Whether or not scientists can bet on the reproducibility of research findings is more than intellectual gaming; the government’s industrial strategy has reproducible science at its core. Coupled with keeping the UK’s research sector thriving after Brexit, this puts more pressure on researchers than ever, but reproducibility needn’t add to that burden. The movement is ultimately about turning the scientific method on itself, and asking: how can we do better?

Marcus Munafò is professor of biological psychology at the University of Bristol and Neil Jacobs is head of open science and research lifecycle at Jisc.