Scientific authorship by gender: trends before and during a global pandemic

The COVID-19 pandemic significantly altered work patterns within the scientific community, including increased work-from-home arrangements, disruptions to lab access, and a shift to online teaching. These changes, coupled with increased domestic responsibilities (childcare, eldercare) disproportionately affecting women, raised concerns about exacerbating existing gender disparities in science. Prior research revealed persistent gender imbalances in scientific authorship, especially in specific disciplines and senior positions. This study hypothesized that the pandemic amplified these disparities, leading to reduced scientific authorship by women due to increased family responsibilities. The study aimed to analyze gender differences in scientific manuscript authorship before and during the pandemic, examining variations across regions and disciplines. It utilized a large dataset of self-identified gender from manuscript submissions to IOP journals, a unique approach compared to previous studies that often relied on less reliable methods like algorithms for gender determination.

Existing literature on the pandemic's impact on gender disparities in scientific authorship presented mixed findings. Some studies indicated higher productivity for men during the pandemic, while others found no significant difference or even higher productivity for women in certain contexts. Methodological differences, including the approach to gender identification (algorithms versus self-reported data), disciplines studied, and measures of scientific productivity (submissions versus publications), contributed to the inconsistencies. The authors critically reviewed 38 prior studies, highlighting the limitations of algorithmic gender identification, particularly for names from certain cultural backgrounds, and emphasizing the need for self-reported gender data.

The study used data on manuscript submissions to 57 IOP journals from January 2019 to July 2021, obtained under a Confidential Information Agreement. The dataset contained the journal, self-identified gender of the corresponding author, and submission date (month and year). The total initial dataset comprised 119,592 submissions. Journals were categorized into seven disciplines based on IOP classifications. The start date of the pandemic for each country was defined as the date with 50 or more confirmed cases. A difference-in-difference approach was used to compare monthly submissions by men and women during and before the pandemic, controlling for country and journal. Data analysis was performed using R and Excel. Regional analysis was based on United Nations regions, and country-level analysis included countries with at least 30 submissions. The analysis focused on the corresponding author's gender as a proxy for scientific productivity.

Of the 119,592 submissions, gender was self-identified for 99,114 (82.9%). Corresponding authors were 82.1% male, 17.8% female, and 0.08% non-binary. Males constituted the majority of authors in all countries and journals. Women contributed 16.5% of articles before the pandemic and 18.8% during the pandemic. While both men and women submitted more manuscripts per month during the pandemic, the pre-pandemic upward trend in submissions slowed for both genders. Although the percentage of female corresponding authors increased during the pandemic, this increase was slower than the pre-pandemic rate, suggesting a potential negative impact. The difference-in-difference analysis, controlling for time, country, and journal, did not reveal a significant shift in gender authorship related to the pandemic. Regional analysis showed similar patterns, with men consistently outnumbering women in all regions, although the proportion of women authors varied across regions. Country-level analysis revealed variations in female representation, with some countries showing higher proportions than others. Disciplinary analysis showed a higher male authorship in astronomy and astrophysics (89.6%) and a lower representation in bioscience (79.0%). In all disciplines, the percentage of female corresponding authors increased during the pandemic compared to the pre-pandemic period, with the largest increase observed in interdisciplinary journals.

The study's findings challenge the initial hypothesis that the pandemic would disproportionately harm women's scientific productivity. While women experienced a higher percentage increase in submissions than men, the pre-existing gender imbalance in scientific authorship persisted throughout the pandemic. The authors acknowledge the limitations of focusing solely on manuscript submissions as a measure of scientific productivity, highlighting other aspects of scientific work (e.g., teaching, mentoring) that could show different gender disparities. The authors discuss several methodological differences between their study and previous research that might account for the differing findings, including sample size, the measure of productivity used (submissions vs. publications), the definition of the pandemic period, the method of determining gender (algorithms vs self-reported), and the disciplines investigated. They compare their study findings to several other studies highlighting the methodological discrepancies and the need for large scale studies utilizing self-reported gender data.

This large-scale study using self-identified gender data reveals persistent gender imbalances in scientific authorship before and during the COVID-19 pandemic, despite a relative increase in submissions by women during the pandemic. While the pandemic may have slowed the pre-existing trend of increasing female authorship, the underlying gender disparity remains a significant challenge. Further research is crucial to investigate the pandemic's impact on other aspects of scientific productivity, considering factors like childcare responsibilities, career stage, and other forms of scientific contributions.

The study's limitations include the lack of information on author age, position, and research experience. Data on self-identified gender were unavailable for 17.8% of participants. The analysis assumes a similar distribution of undisclosed gender in the pre- and pandemic periods, which may be an inaccurate assumption. The limited representation of non-binary authors restricts analysis of this group. The use of month-year submission dates instead of precise dates introduces some uncertainty. Generalizability is limited to the disciplines represented within the IOP journal portfolio and to manuscript submissions rather than accepted publications. The study focused solely on one aspect of scientific productivity; other aspects like teaching and mentoring could exhibit different gender dynamics.