Two-thirds of scientists share results outside their circle of trusted colleagues before formally publishing them in scholarly journals, a survey of thousands of researchers reveals. But sharing trends vary markedly across fields, with social scientists and mathematicians most likely to disclose findings before publication, and computer scientists least likely.

More than 7,000 researchers responded to the survey, the first to ask scientists across disciplines about when and why they share results broadly before publication — for example at conferences or seminars, or as preprints. The respondents held faculty positions in the United States, Germany or Switzerland, and worked in nine different fields — from engineering to biological sciences. Their responses, detailed in Science Advances on 16 May ¹ , reveal that 67% share results before publication.

About 40% of scientists disclose a result after they are sure of its validity, and 21% share findings once they’ve written a manuscript or submitted it to a journal. Only about 6% of scientists disclose preliminary results at the concept stage. Most said they share results before publication to receive feedback; other widely cited reasons included obtaining credit for the work and attracting collaborators.

But whereas nearly 80% of social scientists and mathematicians share results before publication, only about 40% of computer scientists do. The economists who led the study — Marie and Jerry Thursby, who are visiting fellows at Harvard University in Cambridge, Massachusetts — say they didn’t expect the differences to be so large (see ‘Sharing with others’).

Open field

The team found that the tendency to disclose before publication was not associated with the scientists’ gender or age, but with the specific field they worked in. Factors such as competition for funding and commercial relevance of the research explained most of the differences in sharing attitudes. Mathematicians tend to disclose more before publication because they view themselves as being in a fairly non-competitive environment, says Jerry Thursby. By contrast, biomedical scientists perceive their field as competitive, and are less likely to disclose.

Field-specific norms about sharing, such as whether researchers tend to exchange information at seminars and conferences, acknowledge the results of others and get valuable feedback, also influenced how soon scientists disclose their results. “Mathematicians have very good norms when it comes to acknowledging the work of others and giving feedback,” says Jerry Thursby. Researchers in other fields tend to be less open, he adds.

The survey confirms existing assumptions that competition and commercialization influence when scientists share their results, says Eric Campbell, director of research at the University of Colorado’s Center for Bioethics and Humanities in Aurora. But it also shows that the norms that govern a specific field have real implications for when people share their results, he says.

“The core of science is the sharing of knowledge,” says Julia Lane, an economist at New York University. Understanding why and when researchers choose to disclose results is crucial to developing rewards that motivate them to act in the interest of science, she says. “At the end of the day, scientists are people, and people respond to incentives.”