Do You Have to Be Fancy to Get Published in “Science”?

It’s no secret that it’s harder than ever to get published in a top scientific journal.

Faculty remain under the age-old pressure to publish or perish, especially in big-name journals that carry the most influence with tenure and promotion committees, other scientists, and public policymakers. Meanwhile, submissions are at an all-time high, in part because generative AI has helped ramp up the production of papers (of varying quality). One analysis shows that the number of articles indexed in the Web of Science database increased by 47 percent—from 1.7 million to 2.5 million—between 2015 and 2024, far outpacing the growth of the scientific workforce.

That competition also translates to low acceptance rates at top-tier peer-reviewed journals, which range between 5 and 8 percent at Science, Nature, the New England Journal of Medicine and the American Economic Review.

A quick scan of the authors who do manage to get published suggests that a disproportionate share work at selective, wealthy universities.

“When you read articles published in Science or Nature, you see all the famous institutions represented there,” Aaron Clauset, a computer science professor at the University of Colorado, told Inside Higher Ed. “It doesn’t take a high-tech study to see that there’s a correlation between institutional prestige and journal prestige in terms of publishing.”

Prestige Bias?

But little data was available for researchers to independently analyze that metric or any other characteristics of successful authors—until now.

Earlier this year, the American Association for the Advancement of Science gave Clauset and his team unprecedented access to anonymized data from 110,303 manuscripts submitted between 2015 and 2020 to two of its journals, Science and Science Advances. Just as his team sought to “understand how to make the scientific peer-review system more meritocratic by examining where the potential biases might be,” Clauset said, AAAS was “very interested in essentially doing an audit of their own process to understand: Are they being biased? Are there places where they can do better?”

According to the resulting analysis published in Science Advances last week, getting published in Science—which accepted about 6 percent of submissions between 2015 and 2020—or its open-access sister publication Science Advances—which accepted about 10.5 percent of submissions—tends to be easier for scientists with certain characteristics.

For example, manuscripts by authors from the most prestigious institutions had an 11.6 percent acceptance rate, compared to 3.4 percent for those from the least prestigious institutions. (Researchers created an algorithm linked to university rankings to determine prestige levels.)

Although researchers were careful to note that correlation doesn’t equate to causation, the findings raise questions for future analysis: “Is it simply the case that the elite institutions are doing the best work and that the elite journals are just publishing the best work?” Clauset said. “Or is it the case that there may be a prestige bias in the system in which maybe authors from elite institutions are given more of the benefit of the doubt in terms of how their work is evaluated?”

In addition to a possible prestige bias, researchers identified numerous other factors—including research topic, national origin and gender—associated with higher acceptance rates. According to the paper, Science published 9.6 percent of submissions on research focused on viruses, infection, RNA and immune response, but only 1.7 percent of research more focused on how science intersects with politics, economics, gender and the societal dimensions of COVID.

Papers from large teams, with 10 or more authors, were accepted at three times the rate of those produced by small teams, with one to five authors. Papers with authors from the United States or Canada were accepted at 3.3 times the rate of authors based in China. And U.S.-based authors with Chinese last names had lower acceptance rates than their U.S.-based peers without Chinese last names. Women were also at a slight disadvantage, with the journals accepting 5.3 percent of papers written by women compared to 6.4 percent written by men.

At the same time, “bias is difficult to identify separately from quality from these data,” the authors of the study wrote.

“The prestige and team size associations likely reflect a combination of status signaling effects and resource advantages among elite institutions and larger teams that may independently improve manuscript quality,” the study reads. “The high volume of submissions [from authors based in China] may reflect a region-specific incentive to submit work to high-impact journals regardless of suitability, such as China’s past publication bounty program. Editors then rebalance the submission pool to filter these out during editorial review.”

While getting published in a peer-reviewed journal is a multistep process, much of the rejection happens early on. The study found that editors at Science rejected 83 percent of submissions and roughly 75 percent of submissions to Science Advances before ever sending them out for peer review.

However, the qualities associated with higher author success rates “aren’t concentrated at one step of the peer-review process,” said Sam Zhang, an assistant professor of mathematics and statistics at University of Vermont and co-author of the study. “Across the board, both the editors and the reviewers’ [decisions on what to publish] are correlated with these author characteristics.”

As far as Zhang knows, this may be the first time a journal of Science’s influence has released such a large set of manuscript data for independent review. And he’s hopeful that publication of the findings “will be a watershed moment for the publishing ecosystem” and inspire more journals to release similar data. “If every single journal released its manuscript data and one journal’s publications skewed toward one demographic category or another, it would help scientists pinpoint deeper biases.”

And the urgency for influential scientific journals to make sure they’re publishing the best science—not just research affiliated with the biggest names—may be greater than ever in the age of AI.

“The role of elite journals in the ecosystem might become more important as the barrier to entry becomes lower,” Zhang said. “Scientists need some way of filtering the vast amounts of information that they receive. Traditional status symbols like a journal’s reputation are very helpful heuristics for many working scientists to decide what to read.”

As AI increases the broader scientific literature’s volume, he added, “understanding what drives acceptance in these journals might only become more important.”