Publishing in an elite scientific journal is often presented as a straightforward test of research quality. Scientists submit their strongest work, independent experts evaluate it, and the most important studies survive.
A rare analysis of 110,303 submissions to Science and Science Advances shows that the process is considerably more complicated.
The researchers found that papers from prestigious institutions, larger author teams and certain geographic regions achieved substantially higher acceptance rates. Authors based in China experienced markedly lower success, while the overall gender difference was comparatively small.
Most importantly, the strongest disparities appeared before conventional peer review began. Editors rejected more than 70 percent of submissions without sending them to outside referees, meaning a small number of editorial gatekeepers determined which studies were even allowed to compete on the peer-review stage.
The study does not prove intentional discrimination or show that rejected papers were equally strong. It does reveal that publication outcomes at influential journals are connected to far more than reviewer recommendations alone.
The complete peer-reviewed research, titled “Editorial and peer review dynamics at elite general science journals,” is available through Science Advances.
Researchers Opened a Normally Hidden Publishing System
Journal submission records are usually confidential. Outsiders can see published papers but rarely have access to rejected manuscripts, editorial scores or the decisions made at each stage.
That creates a major evidence gap. Acceptance rates are visible, but researchers cannot normally determine whether different groups were rejected immediately, rejected after initial assessment or rejected following full peer review.
For the new study, researchers from the University of Vermont and the University of Colorado Boulder worked with the American Association for the Advancement of Science, which publishes Science and Science Advances.
They examined deidentified records covering 110,303 manuscripts submitted between 2015 and 2020. The data included manuscript topics, author-team size, corresponding-author characteristics, institutional prestige, editorial outcomes and reviewer recommendations.
A public visualization created by the Vermont Complex Systems Institute allows readers to follow thousands of submissions as they move through the editorial system.
Opening this information matters because elite journals do more than distribute scientific findings. Their decisions influence hiring, promotions, funding, media attention and which research questions receive further investment.
Most Papers Never Reached External Peer Review
The popular image of scientific publishing begins with anonymous experts carefully assessing a manuscript.
For most submissions in the dataset, that stage never occurred.
At Science, editors first conducted an initial screening and could reject manuscripts considered unsuitable, insufficiently important or unlikely to interest a broad audience. Promising submissions could then be assessed by the journal’s Board of Reviewing Editors before a smaller group advanced to external peer review.
Of 68,047 submissions included in the detailed Science pathway, only 11,845 reached outside reviewers. Just 4,181 were ultimately accepted, representing approximately 6 percent of all submissions.
Editors rejected 20,755 papers during the earliest stage. Another 35,447 were rejected after the Board of Reviewing Editors assessed them but before full peer review. In total, 63,866 of the 68,047 manuscripts were declined.
Across both journals, about 80 percent of submissions to Science and more than 70 percent submitted to Science Advances were screened out before external peer review.
This does not make editorial screening improper. Highly selective journals receive far more papers than they can publish, and many submissions may clearly fall outside their scope.
It does mean that editors—not outside peer reviewers—perform most of the practical gatekeeping.
Institutional Prestige Was Strongly Linked to Acceptance
One of the clearest patterns involved the corresponding author’s institution.
At Science, manuscripts associated with the most prestigious group of institutions had an acceptance rate of 11.6 percent. Papers from the two least prestigious groups were accepted at approximately 3.4 percent.
That is a difference of more than three to one. A similar general relationship appeared at Science Advances.
Several explanations are possible.
Prestigious universities often have greater funding, stronger laboratories, experienced grant and writing support, and access to international research networks. Their scientists may therefore produce studies that are more expensive, broader in scope or better aligned with elite journals.
Established institutions can also provide a status signal. Editors who recognize a university or laboratory may form expectations about the research before fully evaluating it.
The researchers could not independently measure each manuscript’s underlying quality, novelty or importance. They therefore could not determine how much of the prestige gap reflected stronger research and how much resulted from preferential evaluation.
That limitation is central to interpreting the findings. The study identifies associations, not proof that editors deliberately favored famous universities.
Larger Research Teams Performed Much Better
Team size showed another substantial relationship with publication success.
Manuscripts listing 10 or more authors were approximately three times as likely to be published as papers written by teams of one to five researchers.
Large collaborations can have genuine advantages. They may combine several laboratories, institutions and areas of expertise. They can produce bigger datasets, use expensive facilities and answer questions that would be impossible for one researcher to investigate alone.
A large author list may also indicate that a project belongs to a well-funded and established network.
However, the trend raises questions for fields where small teams or individual scholarship are normal. A theoretical paper, field study or social-science analysis may not require dozens of contributors, yet it competes for limited space against large biomedical or physical-science collaborations.
If team size becomes an informal signal of importance, elite journals may increasingly favor resource-intensive science even when smaller projects make valuable discoveries.
China-Based Authors Faced a Major Disadvantage
Geography also had a strong relationship with publication outcomes.
Manuscripts led by corresponding authors based in China were accepted at less than one-third the rate of papers led by authors in the United States or Canada. The disadvantage appeared at several stages of evaluation and in both journals.
The researchers warned against treating this result as automatic evidence of national bias.
Chinese institutions provide strong professional and financial incentives for researchers to publish in prestigious international journals. That may encourage a particularly high volume of submissions, including papers that are not well matched to a general-science publication.
Language, presentation style, research topic and access to international collaboration may also affect outcomes.
At the same time, geographic reputations could influence how editors interpret a manuscript. Because author identities and affiliations are visible in the journals’ single-blind system, evaluators can see the country and institution before deciding whether the research should advance.
The researchers argued that randomized experiments using double-blind review could help separate differences in manuscript quality from the influence of author identity.
Gender Produced a Smaller Difference
The study found comparatively little overall association between corresponding-author gender and publication success.
At Science, papers led by men showed a small but statistically significant advantage. The effect was much weaker than the differences linked with institutional prestige, geographic location, topic or team size.
The researchers suggested that some of the apparent gender effect may reflect seniority. Senior scientists are still more likely to be men in many fields, and experience can influence the type of projects submitted to highly selective journals.
This finding does not show that gender inequality has disappeared from science. Inequality may affect who receives funding, laboratory leadership, collaboration opportunities and recognition long before a manuscript reaches a journal.
It shows only that, within this particular submission dataset, gender was not the dominant predictor of the final editorial outcome.
Research Topic Could Change a Paper’s Chances Dramatically
The subject of a manuscript was also closely associated with success.
During the early COVID-19 period, studies involving immunology and RNA were nearly six times as likely to be published in Science as social-science research examining the pandemic’s political and societal effects.
Editorial attention naturally follows emerging scientific priorities. When a novel virus threatens global health, research into transmission, immunity and vaccines may have exceptional urgency.
However, topic preferences shape the scientific record. When elite journals favor certain disciplines or approaches, those studies receive more media coverage, citations and institutional recognition.
Research on social behavior, policy or inequality may be highly important but less likely to fit an editor’s interpretation of a major general-science breakthrough.
Editorial selection therefore does not merely identify influential science. It helps determine which science becomes influential.
Reviewers Still Mattered Once a Paper Reached Them
Although editors controlled access to peer review, outside reviewers remained influential after a manuscript reached that stage.
Final editorial decisions correlated strongly with reviewer recommendations. Acceptance rates among peer-reviewed papers were approximately 35.3 percent at Science and 40.8 percent at Science Advances.
This means external review was not ceremonial. Referees could identify weaknesses, request revisions and recommend rejection.
The surprising point is that the demographic and institutional associations were generally stronger during editorial screening than during peer review.
By the time referees assessed a paper, editors had already removed most submissions. Any imbalance introduced at that initial stage could not be corrected later because the excluded work never reached independent experts.
The Study Does Not Prove Elite Journals Are Unfair
The dataset cannot answer the most important counterfactual question: would manuscripts from different institutions or countries receive identical decisions if their scientific quality were exactly the same?
There was no independent and objective quality score against which the editorial decisions could be compared.
Institutional prestige could be associated with resources that genuinely produce stronger research. Larger teams may complete more comprehensive studies. Differences between countries may partly reflect submission volume, topic selection or journal fit.
The authors also caution that analyzing only papers that reached peer review can introduce selection effects. These manuscripts had already survived editorial screening and may therefore differ substantially from the rejected majority.
Accordingly, the findings should not be described as proof of deliberate misconduct by editors or reviewers.
They provide evidence of outcome disparities that require further investigation.
Greater Transparency Could Improve Scientific Publishing
The study’s greatest contribution may be the release of its deidentified data.
Journals frequently ask researchers to share datasets, document methods and make their analyses reproducible. Editorial organizations rarely apply the same transparency standards to their own decision-making.
Regular publication of anonymized submission statistics could help journals identify unexplained disparities involving institution, country, career stage or research field.
Controlled experiments could test double-blind screening, in which editors initially evaluate manuscripts without seeing author names or affiliations. Journals could also review whether editorial boards represent a sufficiently broad range of countries, institutions and disciplines.
The researchers do not argue that every imbalance must disappear. They argue that journals should understand the patterns their systems produce rather than assuming prestige-based publication is automatically meritocratic.
The analysis of 110,303 submissions does not reveal that elite science journals ignore research quality. It reveals that quality is judged through a multistage human process in which status, resources, geography, collaboration size and subject matter are all connected to success.
For scientists, the result explains why rejection from a top journal does not necessarily mean a study lacks value.
For publishers, it delivers a more difficult message: organizations that evaluate the transparency and rigor of science must be willing to examine their own hidden decisions with the same scrutiny.
