Show summary Hide summary
The National Institutes of Health has committed $174 million over five years to a new program called R3PEATS — an acronym for “Rigor, Replicability, and Reproducibility to Promote Excellence, Accuracy, and Translation in Science” — intended to strengthen reproducibility in NIH-funded research. The initiative was announced by NIH Director Jay Bhattacharya in a Science editorial and at an NIH press briefing in September; critics say the plan stops short of directly funding independent replications or enforcing data transparency.
How the money is allocated
The $174 million works out to roughly $35 million a year, or about 0.07% of the NIH’s reported $47.2 billion annual budget. The R3PEATS funds are divided into five categories: about $44.75 million for a “network and community,” $42.4 million for “grassroots rigor projects,” $45 million for “multi-team triangulation,” $38.25 million for “rigor scholar cohort awards,” and $3.75 million for NIH staff and workshops. The plan, as described by NIH, does not earmark any of that money specifically for repeating published studies.
Hegseth visits USS Abraham Lincoln near end of record-long deployment
Pepsi reports drinkers are cutting back and buying less of its soda
Replication costs and what $174 million would buy
Work by nongovernmental projects provides a rough sense of replication costs. The Reproducibility Project: Cancer Biology, supported initially by a $1.3 million grant from the Laura and John Arnold Foundation in 2013, ultimately spent about $50,000 per paper to attempt replications of 50 high-impact experiments. Psychologist Brian Nosek’s Center for Open Science published replications of 164 social science papers this year — about 49% of findings held up — in work funded by a $7.6 million grant from DARPA, which translates to under $50,000 per paper if all funds were devoted to replication. At those rates, $174 million could underwrite a few thousand replications, potentially covering a random sample of several hundred NIH-funded findings per year. But those direct replication activities are not the stated focus of R3PEATS.
Why direct replication was dropped
NIH staff initially proposed creating three to five replication centers that would redo studies selected by a committee. The plan was presented to the Council of Councils, an advisory group made up largely of NIH-funded researchers, and members raised objections. Council participants voiced concerns that a small group might “choose which studies to replicate,” that replications could “re‑adjudicate established science,” and that the effort might “disincentivize innovative research.” Science also reported a related worry that redoing studies could unfairly cast doubt on individual scientists’ work. The objections led NIH to remove the direct replication component from the R3PEATS budget.
Vagueness of goals and metrics
Observers note that the R3PEATS program emphasizes cultural and infrastructural changes — mentorship strategies, methodological knowledge, and building a U.S. reproducibility network — without setting a baseline replication rate, targets, or concrete metrics for success. NIH materials describing the program list anticipated outcomes such as evidence of effective culture-change approaches and a self-sufficient reproducibility network, but do not quantify the size of the reproducibility problem R3PEATS aims to address or provide a timeline for measurable results.
Debate over enforcement versus incentives
Bhattacharya’s Science editorial and NIH communications urge institutions, publishers and funders to act, but stop short of saying the NIH itself will mandate stricter requirements. Critics have argued that the agency could use existing leverage — for example, by requiring preregistration of confirmatory analyses, mandatory sharing of raw data and protocols, making grant renewals contingent on prior-data publication, or auditing a random sample of funded findings and publishing results. Proponents of stricter enforcement point out that many of these measures would cost little compared with the $174 million program but would be unpopular with grantees and therefore require a willing funder to impose them.
Record on data sharing and enforcement
Federal rules and agency expectations on data sharing have a mixed track record. The NIH has told researchers since 2003 that it expects data and protocol information to be shared; in 2016 it added mandatory “rigor and transparency” sections to grant applications. A 2007 federal law requires results of registered clinical trials to be posted within a year, with statutory penalties for noncompliance. A Lancet analysis in 2020 found that only 31% of clinical trials funded by the U.S. government met that posting deadline. According to reporting cited by the NIH and watchdogs, the FDA did not issue its first notice of noncompliance until April 2021 and, as publicly reported, did not collect fines; the agency sent a March 2026 letter reminding roughly 2,200 research sponsors of the requirement. An HHS inspector general review found about half of NIH-funded trials with data due in 2019–2020 reported on time, and the Government Accountability Office reported the NIH did not suspend grant funding for noncompliance before October 2021.
Calls for a different approach
Supporters of stronger enforcement argue that half-measures have failed to prevent reporting lapses and, in some cases, fraud. Reporting by Science’s Charles Piller has documented instances where manipulated figures and incomplete trial reporting were uncovered by unpaid independent sleuths rather than by institutional audits. Advocates say a serious, well-resourced audit program — including random checks and consequences for noncompliance — would change incentives and help protect the scientific record, even as policymakers balance the risks of discouraging legitimate, innovative work.
NIH officials point to tools such as Linked Discoveries, a new browser that color-codes related papers by citation and retraction status, but Bhattacharya has acknowledged such tools “do not determine whether a finding is correct.” The debate now centers on whether R3PEATS’s mix of culture-focused investments and community building will be enough to improve reproducibility, or whether the agency must use its funding power to enforce transparency and fund independent replication directly.
