There is a peculiar circularity at the heart of a new study published this week on MetaArXiv. The paper, by Adrian Barnett of Queensland University of Technology and Matt Spick of the University of Surrey, argues that academic articles accompanied by open peer review reports are significantly less likely to be retracted than those published under traditional closed review. And the study itself, because it was posted as a preprint on MetaROR with its full peer review history visible to anyone who clicks a link, stands as a working demonstration of the transparency it describes. The medium and the message are the same thing.
The researchers examined 116,359 papers published across all PLOS journals since May 2019, when the publisher began offering authors the option to publish peer review reports alongside their accepted manuscripts. They found that roughly 40 percent of authors, 46,920 papers in total, opted in. Among that group, 105 articles were eventually retracted. Among the closed review papers, 321 were retracted. After controlling for a range of confounders using a Cox proportional hazards model, the association held: papers whose authors chose transparency were less likely to be pulled from the literature.
The finding is not causal, and Barnett and Spick are careful to say so. This was an observational study, not a randomized trial. The preferred explanation runs in the opposite direction from what a reader might assume. It is not that making peer review reports public somehow protects a paper from later retraction. Rather, researchers who are already confident in the rigor of their work and who embrace open science practices tend to select into the open review group. The decision to opt in is itself a signal of article quality, the authors suggest, a kind of voluntary audit that confident researchers are willing to submit to.
To test whether authors might be cherry picking only favorable reviews, Spick conducted a follow up sentiment analysis comparing reviewer language in PLOS journals with that in BMC journals, which also publish open reports. He found no difference in sentiment between the two sets of reviews. That result undermines the idea that the lower retraction rate is simply a selection artifact of authors choosing to publish only their most flattering peer reviews.
The study draws on a dataset built by downloading the full PLOS corpus in XML format on 1 March 2026. The researchers excluded non peer reviewed content such as editorials and obituaries, and also excluded any articles accepted before the policy change took effect on 22 May 2019. Citation data came from the OpenAlex database, and retractions were traced through CrossRef and PLOS metadata.
National patterns in opting in
One of the more striking findings concerns geography. Authors based in the United Kingdom, the Netherlands, France, and Ethiopia were most likely to opt for open peer review. Those in Saudi Arabia, South Korea, Pakistan, Poland, and China were least likely. Cultural norms around academic debate, funder encouragement, and national publishing mandates all appear to shape these decisions. Major funders such as the Wellcome Trust in London and the Gates Foundation actively promote open science practices, and their influence shows up in the data.
This geographic variation complicates the picture. If openness is a marker of confidence and quality, but the willingness to be open varies by country and by funding environment, then the relationship between peer review transparency and research integrity is partly a product of institutional infrastructure and incentives, not just individual author behavior.
The study as its own evidence
The preprint was submitted to MetaROR, a platform that publishes peer reviewed research alongside the full records of the review process. Any reader can examine the reviewer comments, the authors’ responses, and the editorial decisions that led to the preprint being accepted. The openness that Barnett and Spick studied in the PLOS corpus is the same openness that made their own paper’s evaluation visible. The reader who wants to check whether the reviewer raised concerns about causal interpretation can do so in real time. One reviewer, Reese Richardson, suggested expanding the analysis to include expressions of concern, a related but distinct category of post publication notice. Another, Jiang Li, questioned whether the study’s design could cleanly separate author confidence from article quality, and pushed back on some of the causal language in earlier drafts. Neither comment is hidden. Both are published alongside the paper.
That level of transparency is still unusual in academic publishing. The standard model, still dominant across most journals, sends a manuscript to two or three anonymous reviewers whose identities are concealed from the authors and whose reports are shared only with the editor. The authors see the feedback but the wider field does not. Open peer review flips that arrangement: the reports are published, sometimes with the reviewer’s name attached, sometimes without. PLOS was an early adopter of the opt in model, which is why it provides such a large natural experiment for this kind of study.
Caveats and limitations
The association between open review and lower retraction rates is statistically robust, but the authors list several important limitations. Retractions are rare events in absolute terms, which means the statistical models are working with small numbers. The data do not distinguish between retractions for honest error and retractions for fraud or misconduct, two very different categories with very different implications for any argument about quality. And again, the central caveat: correlation is not causation. The authors who choose open review might be systematically different from those who do not in ways that are hard to measure.
The Nature News coverage by Dalmeet Singh Chawla, published the same day, notes all of these limitations. The news article observes that the study cannot answer whether making peer review reports public would reduce retractions if it were mandated, since the whole analysis is based on voluntary opt in. The authors of the Nature piece do not weigh in on this either.
What the finding means
Still, the study adds a substantial piece of evidence to a long running debate. The argument for open peer review has traditionally rested on principle: science is supposed to be self correcting, and hiding the corrective process from view seems inconsistent with that ideal. Critics have worried that open review might make reviewers less candid, or that junior researchers would be reluctant to criticize senior figures if their names were attached. This study does not address those concerns directly, but it gives proponents a new empirical argument. However the mechanism works, papers whose authors are willing to be transparent about the review process end up being more durable in the literature.
For readers who want to judge the evidence for themselves, the preprint and its review history are freely available at MetaArXiv under DOI 10.31222/osf.io/2b7z5_v1. The Nature News article by Dalmeet Singh Chawla is at DOI 10.1038/d41586-026-02282-1. Neither requires a subscription. Both, in their own way, practice the openness the study describes.
References
Barnett, A. & Spick, M. (2026). Who chooses open peer review and is it an indicator of article quality? An observational study of PLOS journals. Preprint at MetaArXiv. DOI: 10.31222/osf.io/2b7z5_v1
Chawla, D. S. (2026). Papers with open peer review less likely to be retracted. Nature News. DOI: 10.1038/d41586-026-02282-1