An independent institute focused on improving the replicability of UK scientific research should be created as part of an overhaul of science structures and funding, a thinktank has recommended.
Calling for a new “National Institute for Scientific Replicability”, the Social Market Foundation says the proposed centre would tackle the “crisis of replicability” that has hit many scientific disciplines, with researchers following up high-profile findings unable to confirm initial results.
The lack of replicability may reflect growing pressure on researchers to publish “only positive results”, which has led to dubious methods being used to obtain eye-catching discoveries, says the report, published on 31 July.
“Because only positive results create the hype and the prestige required for academic career progression, researchers typically don’t publish null (negative) results, creating publication bias,” explains the report’s author Will Henshall, a researcher at the Harvard Kennedy School.
“This creates inefficiency if researchers try the same experiments, being unaware of the null result obtained by other scientists, and if enough negative results are left in the file drawer instead of being published, false claims can become canonised as fact.”
According to the non-partisan thinktank, the new replicability centre should be independent from other scientific bodies and have its own budget.
It should also be given the power and resources to fund new lines of enquiry on important scientific questions where the traditional sources of funding are arguably affected by “groupthink and confirmation bias”.
The institute should also be tasked with sponsoring independent replication projects and developing tools that can predict the replicability of scientific enquiries to guide replication efforts, says the report.
The report also raises concerns that the productivity of scientists in the UK has been falling in recent years, having declined particularly sharply since the 1990s. As a result, the paper argues that taxpayers are “getting less back for our buck” despite the number of researchers rising over time.
To address these issues the report recommends that greater efforts should be made to share research infrastructure, including the creation of repositories of preprints, data and code that has not appeared in a peer-reviewed journal, and more public datasets in the vein of the UK Biobank and the NHS Genomic Medicine Service.
The study also calls for UK Research and Innovation to invest more in outstanding individuals, rather than projects, via fellowships for promising young researchers, while at least 5 per cent of UKRI grants should be spent on “radical ideas”, with reviewers given a “golden ticket” to fund any project that they think is particularly promising, regardless of the views of other reviewers.
The thinktank report also calls for trials of new funding mechanisms such as assigning funding by lottery and backs the creation of an “Atlas Institute”, which would encourage scientific productivity and interdisciplinary research – an idea previously proposed by the Tony Blair Institute.
