Science Press Releases, Not Social Media, Decide Which Studies Make the News

For two years, the communication office at the Institut de Neurociències de la Universitat Autònoma de Barcelona (INc-UAB) ran what amounted to a live experiment on its own scientists. Every time one of its 61 principal investigators published a paper, staff logged which promotional path it took: a press release, a social media post, or nothing at all. They then tracked what happened next in the news and online for up to three months. The results, published in Scientometrics, show a strikingly narrow gate controlling which findings the public ever hears about.

Among 201 original research articles produced by the center between 2016 and 2021, only 28 were accompanied by a press release. The other 173 got no dedicated media push. When the team compared the two groups using Altmetric, a platform that tracks online mentions of research, the gap was not subtle: press-released papers picked up seven news mentions on average, while the rest picked up fewer than one-tenth of a mention.

The retrospective count was suggestive but not conclusive, since papers chosen for a press release might simply be more newsworthy to begin with, a confound the researchers openly acknowledge. So the team ran a second, prospective experiment: they selected 25 upcoming papers and deliberately assigned some to receive a press release and others not to, then measured each paper’s own attention metrics before and after the intervention, using each article as its own control. Papers that got a press release began accumulating mentions in the news within hours; the ones that did not received zero, even three months later. The design let the researchers isolate cause from mere correlation in a way that earlier observational work on environmental health papers could only gesture toward.

Social media told a different story entirely. Posting about a paper on X, whether from the institution’s own account or elsewhere, showed almost no relationship to whether journalists picked it up. Among the 173 papers that never got a press release, only nine drew any news coverage at all, and the number of tweets about a paper had no measurable link to its odds of being reported once journal impact factor was accounted for.

“Our study shows that journalists are not the only gatekeepers determining which scientific findings reach the public,” said Roser Bastida, the paper’s first author and a doctoral graduate of the Universitat de Vic – Central University of Catalonia. “Research institutions themselves act as a first filter, because studies without a press release are likely to go unnoticed, no matter how relevant they may be. This means that research produced by institutions without communication departments will have a much more limited public reach.”

The composite Altmetric Attention Score (AAS), which folds news mentions, tweets, and other signals into a single figure, captured the same pattern at a larger scale. Press-released papers averaged an AAS of 76.3; papers without one averaged 4.58, a 16.6-fold gap that held steady across every journal-quality bracket the team tested, from modest journals to some with impact factors above 10.

In the prospective arm, papers with a press release gained 68.3 AAS points in the first week after release, compared with 0.6 points for papers monitored without one. That comparison undercuts a common assumption in the field, since Altmetric’s own documentation stresses that its score is a weighted count of attention, not a measure of research quality, meaning the gap traces back entirely to promotion rather than to any property of the papers themselves.

The team also found that Altmetric itself was missing a large share of the attention it claims to track. Using manual Google searches, the researchers turned up 106 additional news mentions of press-released articles that Altmetric’s automated system had failed to detect, 63 percent of the true total. On X, undetected posts made up 82 percent of all mentions of press-released papers. Much of that gap traced to a simple technical cause: Altmetric only counts a mention if it links directly to the publication or its DOI, so a news story or tweet that merely describes a finding without citing the source paper slips through unrecorded.

One further wrinkle complicated the tidy story: timing seemed to matter less than expected, as long as institutions stayed within a fixed window. Once 22 days had passed since a paper’s publication, delaying a press release further made no measurable difference to how much attention it eventually drew, a finding the authors link to prior research on how quickly online mentions of new papers typically accumulate. The one hard constraint came from EurekAlert!, the platform many press offices use to distribute releases internationally, which will only accept material describing findings first made public within the previous 90 days.

Sample size is the study’s clearest limitation: the prospective experiment involved just 25 papers from a single mid-sized neuroscience center, and the authors caution that findings from one field and one institution may not transfer cleanly elsewhere. A single case in the data makes the point vividly anyway. Two papers with unusually high impact factors, 25 and 16.6, ran through opposite campaigns purely because of author preference. The one with a press release drew 28 news stories and 315 X mentions within a week, reaching an Altmetric score of 219. The one without gained a single AAS point in three months.

Salva Ferré, a science communicator and one of the study’s coordinators, argues the findings should reshape how funders think about measuring research’s public value. “The findings highlight the importance of press releases while raising questions about the effectiveness of other science communication activities, particularly on social media,” he said. “At the same time, they underline the need to rethink how the societal impact of research should be assessed, as this is becoming an increasingly important criterion in research funding calls.”

What the data can’t settle is which papers deserve that push in the first place. The team’s own decision tree for choosing which of the 25 prospective articles to promote leaned on staff judgment about general interest, exactly the kind of subjective call the study set out to route around for the press release itself. Institutions without a communication department, and researchers whose work never crosses a press officer’s desk, remain outside the system this study describes entirely, whatever the merits of their findings.

FAQ