Low representation of women among authors of retracted papers more likely reflects the structural imbalances of scientific systems than the intrinsic quality of the work of male and female scientists, according to the author of a new study.

Paul Sebö, of the University Institute for Primary Care at the University of Geneva, analysed 878 retracted papers that were published in 131 “high-impact medical journals”, using the artificial intelligence tool Gender API to guess the gender of the papers’ authors based on their names.

Authors assumed to be women, Sebö concluded, represented only 16.5 per cent of first authors and 12.7 per cent of last authors of the retracted publications; a “marked” under-representation, he said, because according to prior research, women make up 45 per cent of first authors and up to 33 per cent of last authors in “comparable” publications.

Women were also under-represented among authors with multiple retractions, Sebö found. Of the 37 authors with repeated retractions identified, only three were assumed to be women. The study also indicated a more general gender imbalance, with women comprising 23.1 per cent of authors across the papers analysed.

51�Թ�

Even though “women are under-represented in retractions relative to their presence among biomedical authors”, Sebö told 51�Թ�, “that does not mean women are less likely to commit error or misconduct”. Rather, the study findings are likely to indicate a broader gender disparity in the field of biomedical science.

“Retractions disproportionately involve senior authorship and laboratory leadership roles, and women remain under-represented in those roles across most medical specialties,” Sebö told 51�Թ�.

51�Թ�

“The simplest explanation for the pattern we observe is structural: if men occupy more senior positions and lead more research groups, they are naturally more exposed to retractions, whether for honest mistakes or for misconduct.”

The findings, he summarised, “reflect who holds responsibility in the scientific system, not the intrinsic quality of men’s or women’s science”.

In general, retractions are a “particularly informative lens for studying inequalities in science”, Sebö said. “Retractions are more than corrections to the scientific record” because they “reflect how responsibility, authorship and leadership are distributed within research teams”.

“By examining who appears in retractions, we obtain an indirect view of how scientific power and accountability are structured across the research workforce,” he said.

51�Թ�

“Retractions highlight how unevenly scientific responsibility and opportunity are distributed, and efforts to strengthen research integrity will be most effective if they take those structural inequalities into account.”

Sebö acknowledged several limitations to his study, noting that Gender API can fail to correctly guess authors’ gender and fails to account for non-binary identities. Further work, he said, should involve qualitative interviews with authors of retracted papers, enabling a greater understanding of “how gender might shape experiences around investigation and retraction”.

emily.dixon@timeshighereducation.com