A UVM Researcher Opened a Window Into How Top Science Journals Decide What to Publish
Sam Zhang favors “double-anonymous” review, in which reviewers cannot see who the authors are, so that the effect of a name and affiliation can be measured directly.
The gatekeeping that picks winning papers usually happens out of sight. Sam Zhang's team got inside two of the world's leading journals — and he is now urging other publishers, including in Vermont, to reveal how their own gatekeeping works.
Getting a paper into one of the world’s most prestigious science journals can shape a career, steer funding, and help decide which discoveries the public ever hears about. Yet the process that picks the winners happens almost entirely out of view. Manuscripts go in; acceptances and rejections come out; what happens in between is rarely visible to anyone outside a journal’s editorial office.
A new study led out of the University of Vermont is one of the first to get inside that process. Its finding: papers from large research teams and elite universities are accepted at higher rates than those from smaller teams and less prominent institutions — a gap the researchers are careful to say is not, on its own, evidence of bias. And its lead author is now asking why more of science’s gatekeepers don’t let outsiders examine their work the way these two journals did.
Sam Zhang, an assistant professor of mathematics and statistics who directs the Science and Humanity Lab at UVM’s Vermont Complex Systems Institute, and colleagues at the University of Colorado Boulder assembled an anonymized dataset of more than 110,000 papers submitted to Science and Science Advances between 2016 and 2020. The analysis was published in Science Advances.
“We model the human side of science,” Zhang said of his lab — studying both “the human infrastructure that powers the scientific enterprise” and “what it means to carry human values into the practice of science.”
What the numbers show
Across the submissions studied, the researchers found that papers from larger author teams, from higher-ranked institutions, and from certain regions were accepted at higher rates. Manuscripts from the most prestigious institutions were accepted at roughly 11.6 percent, compared with about 3.4 percent from the least prestigious. Larger research teams were roughly three times more likely to be published than the smallest. Corresponding authors based in China were more than three times less likely to be published than those in the United States and Canada.
Zhang is emphatic that those gaps are associations, not evidence of bias — a distinction he returns to when asked what he would not want a reader to conclude.
“The reason that we can’t say it’s bias is because there can be genuine differences in the types of science being done across these characteristics,” he said. Smaller research groups tend to work on different kinds of problems than large ones, he noted, and elite institutions “have more resources such as students and funding as well.” Whether the disparities reflect the manuscripts themselves or something about how authors’ names and affiliations are perceived, he said, is a question the data alone cannot settle.
How the data got out — and a disclosure
Opening a journal’s submission records to outside researchers is unusual. Zhang credited “the forward-thinking nature of the editorial leadership at Science,” and pointed to a second factor: the study’s senior author, CU Boulder’s Aaron Clauset, is also a deputy editor at Science Advances, which he said gave the team a chance to demonstrate it could handle sensitive data responsibly. The submissions were anonymized before analysis.
Zhang offered that arrangement without being asked, and the disclosure is worth stating plainly for readers weighing the findings: one of the study’s authors holds an editorial position at one of the journals under examination.
The China figure, with a caveat
Zhang was measured about the regional gap, which is the study’s most striking single number on its face. He declined to attribute it to any one cause and instead flagged a limit of the data itself: the submissions run only through 2020, the year China moved to end cash rewards for publishing in certain journals. Whether the pattern has held since, he said, is a question for future work rather than a settled finding.
Not an alarm
Zhang connects the study to a strain the field is already feeling. But asked directly whether he is worried about scientific publishing’s ability to keep up, he was not.
“There’s already more science being published than any single person can read,” he said, noting that this predates AI. Elite journals, in his view, serve a filtering function that becomes more valuable as the volume grows. And the enterprise has adapted before: he pointed readers to the work of historian Melinda Baldwin, of the University of Maryland, who has documented how recently — and how unevenly — formal peer review actually took hold at scientific journals.
What he would like to see is experimentation. Zhang favors “double-anonymous” review, in which reviewers cannot see who the authors are, run as a controlled trial on a subset of manuscripts so that the effect of a name and affiliation can be measured directly. Such review already happens in some fields but is not the norm at interdisciplinary journals like Science. The obstacles, he said, are inertia, cost, and the need for buy-in from editors and authors alike — real but, in his words, surmountable.
A challenge to the gatekeepers
Asked what he hoped the study would prompt, Zhang aimed it squarely at the journals and publishers themselves — not at authors or their universities — over how much of their own inner workings to make public.
He said he hopes it is “a watershed moment for publishers around the country, including Vermont, in thinking about the trade-offs they’re making between confidentiality and transparency, and whether they can afford to do something like this too.” What Science did, in other words: release its own submission and review records, stripped of identifying details, so that outsiders can study how the gatekeeping actually works.
The study, “Editorial and peer review dynamics at elite general science journals,” was published in Science Advances on July 15, 2026 (DOI: 10.1126/sciadv.aec0494). Sam Zhang responded to Compass Vermont’s questions in writing.